Professional Documents
Culture Documents
September 2007
Abstract
This Working Paper should not be reported as representing the views of the IMF.
The views expressed in this Working Paper are those of the author(s) and do not necessarily represent
those of the IMF or IMF policy. Working Papers describe research in progress by the author(s) and are
published to elicit comments and to further debate.
The paper assesses the financial situation of the health sector in the Slovak Republic. It also
evaluates the efficiency of health expenditures and service delivery in comparison to the
OECD and other new EU member states and suggests avenues for cost recovery and reform.
The health sector of the Slovak Republic is plagued by financial problems. To turn around
health system finances and achieve larger gains in health outcomes, the efficiency of health
spending needs to increase and the mix and quality of real health resources need to be
improved. Although Slovak’s overall health spending efficiency is on par with that of the
OECD, substantial inefficiencies occur in the process of transforming intermediate health
inputs into health outcomes. Efficiency may be enhanced by containing the cost of drugs and
reducing reliance on hospital care. Also, although cost-effectiveness may be relatively high at
present, its sustainability in the future is an issue.
1
This paper was initially written and published as a Selected Issues Paper (SM/07/182). The authors gratefully
acknowledge the comments of Biswajit Banerjee and Engin Dalgic.
2
Contents Page
I. Introduction ............................................................................................................................3
II. Recent Health Reforms and the Fiscal Challenge of Health Care ........................................3
References................................................................................................................................24
Tables
1. Total and Public Health Expenditures, 2000–07 ...................................................................3
2. Health System Outstanding Debt and Arrears, 2004–06 .......................................................4
3. Public Spending on Health Care, 2005–10 ............................................................................6
4. Selected Real Health Resources.............................................................................................8
5. Expenditure on Pharmaceuticals, 1999–2002........................................................................9
6. Health Outcomes in the OECD and NMS-10 ......................................................................10
7. Output-Oriented Efficiency .................................................................................................13
8. Rank of Health Efficiency Scores Relative to the OECD....................................................13
9. Correlations of Relative Efficiency with Associated Factors ..............................................16
Figures
1. Health Expenditure in the OECD and NMS-10, 2000–04.....................................................7
2. The Efficiency Relationship Between Health Expenditure, Resources, and Outcomes......10
3. Spending to Outcome Frontier HALE .................................................................................12
4. Efficiency and the Best-Practice Frontier ............................................................................21
Appendix
I. Technical Methodology........................................................................................................20
3
I. INTRODUCTION
The Slovak Republic finds itself at a crossroads in health care reform. The current
government has rolled back the main measures of the reform package put in place during
2003–04 in the face of public disenchantment with the results of those reforms. But
further change is needed to raise the efficiency of health spending, so that the health
system can be put on financially sound footing and health outcomes can be improved.
The current government has formulated some initiatives, but is still in the process of
developing a comprehensive strategy for the health sector.
This paper provides an analysis of key issues in the health sector and recommendations
for a health reform strategy. Section II focuses on recent reforms and the fiscal challenges
in the health care system. The conclusion of this analysis is that a well-defined strategy is
needed to control the fiscal cost of health care over the medium term. In Section III, we
turn to the question of the efficiency of health spending—a key issue for controlling
health care cost and improving health outcomes. The main finding is that an immediate
challenge for the Slovak health care system is to improve the mix of health care resources
(e.g., doctors, hospital beds, and pharmaceuticals). The analysis also suggests that more
attention should be paid to pharmaceutical costs, doctors’ consultations, bed utilization,
and outpatient contacts. Finally, in Section IV, we present some recommendations on
measures that could be part of a health reform strategy. This includes strengthening of
central oversight over public hospital finances; enhancing the role of the private sector;
and reforming financial arrangements.
II. RECENT HEALTH REFORMS AND THE FISCAL CHALLENGE OF HEALTH CARE
(percent of GDP)
Total 1.4 0.4 0.4 0.4 0.4
Health institutions 1.3 0.4 0.4 0.3 0.4
Ministry of Health 1.0 0.2 0.3 0.1 0.2
Regional and local governments 0.2 0.2 0.1 0.2 0.1
Insurance companies 0.2 0.1 0.0 0.1 0.0
The reform measures were not sufficiently strong to resolve the financial problems of the
health sector. The co-payments were relatively small (Sk20 for doctor visits and Sk50 per
day of hospital stay). Because the coverage of the mandatory health insurance was left at
very high levels, the demand for the newly introduced supplementary voluntary health
insurance was low. The change in the legal status of hospitals fell short of efforts to
privatize hospitals. Thus, public health spending started to rise again in 2005, and large
state-owned and regional hospitals continued to accumulate new arrears with their
suppliers, particularly pharmaceutical companies. The reforms also were very unpopular.
Public opinion polls revealed widespread disapproval with the health reforms,2 and health
care policy was an important issue during the 2006 electoral campaign.
The new government that assumed office in June 2006 reversed key elements of the
2003–04 health care reform. Co-payments for doctor visits and hospital stays were
abolished and co-payments for drugs were lowered significantly; profits and
administrative spending of health insurance companies were limited to 4 percent of their
total expenditure; and legislation was submitted to parliament to change the legal status of
the state-owned insurance companies from joint stock companies to public agencies.
The government also undertook measures aimed at bolstering the finances of health
insurers and health care institutions, but a comprehensive reform strategy remains to be
formulated. With the aim of reducing the cost of medical services, the Value-Added Tax
(VAT) rate for most pharmaceuticals was reduced from 19 percent to 10 percent. In
addition, the government increased the transfers to health insurance companies for health
insurance contributions to cover the non-working population (e.g., pensioners and the
unemployed). For the first quarter of 2007, these transfers were raised from 4 percent to
2
An opinion poll found that 74 percent of respondents disagreed with the introduction of the health care
reforms, compared with 35 percent of the respondents opposing pension reform (Jevčák, 2006).
5
5 percent of the minimum wage per insured person. This rate would revert back to
4 percent thereafter, unless the health ministry formulates a plan for lowering the fiscal
burden of health care. The health ministry has identified some 6,200 hospital beds (about
16 percent of the total number of beds) that it deems redundant and that should be
eliminated. However, the implementation of the targeted reductions depends on the
collaboration of subnational governments (SNG), which control most of the hospitals
with the identified beds.
Further reforms will be needed to enable the health care system to remain within the
financial envelope specified in the 2007–09 budget framework. At current rates,3 health
insurance contributions and other public resources available for spending in the health
sector are projected to decline from 5.3 percent of GDP in 2007 to 5.1 percent during
2008–10 (Table 3). There is a risk that wage increases and the rising cost of
pharmaceuticals4 will crowd out other health spending. This would jeopardize the quality
of health services and likely result in additional arrears accumulation. Managing these
pressures will require the implementation of reforms aimed at raising the efficiency of
health spending.
Total health spending in the Slovak Republic is less than one-third of the
EU–15 and OECD averages and above the median for the NMS–10. This partly reflects
the higher cost of health services and increased health care demand in countries with
higher income levels. The share of the private sector in total health care spending in the
Slovak Republic is among the lowest in the EU (Figure 1). Average annual per capita
expenditure on health care in the Slovak Republic during 2000–04 in purchasing power
parity (PPP) dollars was PPP$712, of which only 11 percent came from private sources.6
3
See Health Policy Institute (2007) which is consistent with government forecasts.
4
After declining slightly in 2007, owing to the reduction in the VAT rate for most pharmaceuticals, the cost
of pharmaceuticals is likely to resume its upward trend from 2008 onward, in line with envisaged trends of
international pharmaceutical prices. Moreover, the recent increase in drug costs is due in part to a trend
toward using a larger volume of more effective, but more expensive medicines (Nemec and Ochrana, 2005).
5
The sample of countries also include the new EU-member states Cyprus and Malta.
6
Spending is measured in PPP terms in order to be able to compare expenditure levels across countries.
More conventional measures of spending would bias such a comparison. For example, spending measured
as a percent of GDP underestimates the purchasing power of spending in richer countries relative to poorer
countries (because a comparable package of health services will cost less as a percent of GDP in the richer
country). At the same time, richer countries should be expected to spend more on health care in PPP terms;
6
This reflects the high coverage of the mandatory public health insurance which leaves
little space for private supplementary insurance. Only the Czech Republic has lower
private health spending, at 9 percent of total health expenditure.
Estimate Projection
2005 2006 2007 2008 2009 2010
as populations grow wealthier, they are likely to consume a larger and more varied package of social
services, leading to increased spending (the Wagner effect).
7
2500
1000
500
e
lic
ia
nd
a
li c
ge
ia
ge
ia
ia
ry
ag
ni
ni
an
ar
ub
en
tv
ub
ga
la
ra
ra
to
ua
er
lg
La
om
Po
ep
ov
ve
ve
ep
un
Es
th
Av
Bu
R
Sl
A
R
R
Li
D
ak
10
5
ch
EC
-1
ov
S-
ze
EU
O
M
Sl
N
Public health expenditures Private health expenditures
Compared to other countries, public health care expenditures and resources are tilted
toward hospital care, pharmaceuticals, and wages. Hospital bed availability is in line with
the NMS-10 average, but higher than the averages of the EU-15 and OECD. However,
the hospital bed occupancy rate and the rate of inpatient care admission are lower in the
Slovak Republic than in comparable countries (Table 4). On the other hand, the use of
outpatient and doctors’ services is high compared to other countries. Spending on
pharmaceuticals is also higher in the Slovak Republic than in the other NMS-10 countries
(Table 5). Spending is also biased toward compensation of employees, which amounted
to 44.6 percent of total spending by health facilities in 2004, as compared to an average of
27.7 percent in the EU-15 (Institute for Health Information and Statistics, 2005, and
Eurostat Task Force on COFOG, 2006). Overstaffing of physicians and accompanying
health staff in relation to EU-15 appears to be a key issue (OECD, 2006). While wage
levels are low, pressure from unions for wage increases is rising and there is anecdotal
evidence of health staff emigration to EU-15 countries.
Table 4. Selected Real Health Resources 1/
NMS-10 average 7.0 3.0 9.8 45.4 10.9 72.7 21.2 8.6 8.8 97.0
EU-15 average 5.5 3.2 13.0 82.5 5.9 74.3 17.9 8.4 5.4 89.7
OECD average 6.1 3.0 12.5 74.4 6.9 76.2 18.6 8.4 7.0 91.8
Sources: WHO and the World Bank’s World Development Indicators database.
1/ Data are from latest year available except for the data on doctors’ consultations, which are the average over 2002–03, and immunization, which
are from 2004.
9
Health outcomes in the Slovak Republic are close to the average for the NMS-10 but
significantly worse than the average for the EU-15 and OECD. According to the latest
available data, health adjusted life expectancy (HALE) in the Slovak Republic is 66
years, five years less than the EU-15 and OECD averages (Table 6). Death rates
(standardized by population demographics), infant and child mortality rates, and the
incidence of tuberculosis are also worse. However, maternal mortality rates are relatively
low in the Slovak Republic compared to other NMS-10 countries and comparable to
EU-15 and OECD averages.
Efficiency analysis assesses whether expenditures are higher than needed to achieve
prevailing health outcomes. Like other NMS-10 countries, the Slovak Republic combines
relatively low health spending with relatively poor health outcomes. However, by
increasing expenditure efficiency, it may be possible to raise health outcomes without
increasing spending or, vice versa, to reduce spending without compromising outcomes.
Figure 2 illustrates the concept of spending efficiency. Overall spending efficiency links
health expenditure with health outcomes. The link between spending and health outcomes
can be broken down into two stages. The first stage measures cost effectiveness—i.e., the
efficiency of spending on intermediate outputs or real health resources, such as hospital
beds, number of health workers, etc. The second stage measures system efficiency—
i.e., how well the intermediate outputs or real resources are used to achieve health
outcomes such as improved life expectancy and lower mortality rates.
10
Overall
efficiency
Cost System
effectiveness efficiency
11
The Slovak Republic’s overall spending efficiency is on par with that of OECD countries
and other NMS-10 countries. On average, the Slovak Republic ranks in the 54th percentile
of the efficiency score ranking of OECD and NMS countries for public health
expenditure (Table 7). If private health expenditures, which are relatively low in the
Slovak Republic, are taken into account, the Slovak Republic ranks higher, at the
22nd percentile, in the efficiency score ranking for total spending on health. The Slovak
Republic’s ranking indicates that there is scope for improving outcomes without
increasing spending.
Inefficiencies in the Slovak health system occur mostly in the process of transforming
intermediate health resources into health outcomes (Table 8). In other words, system
efficiency is relatively low in the Slovak Republic (see Figure 2 and Appendix I). 10 This
reflects a general feature of NMS-10 countries, which achieve relatively low health
outcomes with high real resource combinations. In part, this is due to inertia—for
instance, hospital structures may still reflect old standards and a significant number of
current health workers were educated in the pre- and early transition period. On the other
hand, higher levels of cost effectiveness (see Appendix I) in the Slovak Republic and
NMS-10 countries reflect relatively low prices for labor and other inputs for health
services. As a result, despite spending levels, real resources in the health sector are
relatively high.
7
The sample of countries included in the analysis are OECD countries (except Mexico and Turkey, as their
levels of health outcomes and spending make them outliers) and the EU new-member states of Bulgaria,
Cyprus, Estonia, Latvia, Lithuania, Malta, Romania, and Slovenia. Croatia was also included in the
analysis, although some data were not available.
8
The methodology derives from the literature on the estimation of production functions. DEA has the
advantage of being sparse in its assumptions about the characteristics of the production technology. This is
particularly important for assessing spending efficiency, because little is known about the nature of the
relationship between spending, intermediate outputs, and outcomes.
9
By using average health expenditures over 2000–03 and health outcomes in 2002 and 2004 in the DEA,
we allow for a time lag between when spending takes place and when health outcomes are measured. The
exceptions are maternal mortality, where the latest outcome data available are for 2000, and standardized
death rates, where two countries have data available only for 2001.
10
To assess at what stage in the process inefficiencies arise, the individual countries’ average rankings of
overall expenditure efficiency were compared with the average rankings of their efficiency in transforming
intermediate resources into outcomes.
Figure 3. Spending to Outcome Frontier HALE
75 Japan
Sweden
Switzerland
Spain ItalyAustralia Iceland
Austria Canada FranceGermany Norway
Netherlands
Malta Finland Belgium Luxembourg
Greece New Zealand United Kingdom
70 Slovenia Ireland Denmark
Portugal United States
Korea Czech Republic
Cyprus
HALE
Croatia
Slovak Republic
Poland
65
12
Bulgaria Hungary
Estonia
Lithuania
Romania
Latvia
60
0 500 1000 1500 2000 2500 3000 3500
Public health expenditures (PPP per capita)
Percentile
1-25 26-50 51-75 76-100
Public expenditure Bulgaria Czech Republic Estonia Hungary
Latvia Poland Lithuania
Slovak Republic Romania
Slovenia
modernized system of health care would likely require substantially higher spending
levels, for example for reorganizing hospital care and employing high-quality health
workers.
These results are broadly consistent with the findings of other studies, although
methodologies and data differ. A study of public sector efficiency in Czech Republic
ranks the Slovak Republic around the average of NMS-10 countries for overall input-
oriented health efficiency but substantially lower in converting real health resources into
outcomes (IMF, 2007). Furthermore, similar work for Slovenia ranks the Slovak Republic
among the worst of the sample of 22 OECD and other NMS countries, although this study
uses public-only health spending in percent of GDP as the input (Mattina and
Gunnarsson, 2007). Afonso and St. Aubyn (2007) rank the Slovak Republic and other
NMS countries in the bottom third of the efficiency distribution of a wide sample of
countries using output-oriented overall health efficiency scores, and second-to-last when
assessing system efficiency.
C. Correlation Analysis
It is important to understand the reasons for differences in relative efficiency between the
Slovak Republic and comparable countries. Many policy-related factors and factors out of
the direct control of policy makers (environmental variables) affect the relationship
between health expenditures and health outcomes. We examine what factors determine
the variation in the link between health spending and outcomes across countries by
simultaneous multi-correlation analysis.11 Lessons are drawn about which policy factors
are important to consider for improving health sector efficiency in the Slovak Republic.
Efficiency is associated with a wide range of factors. This is summarized in Table 9. GDP
per capita is highly and negatively correlated with overall relative efficiency, reflecting
changes in relative prices of health care as income increase (see footnote 4).12 Because of
the pervasive impact of GDP, all reported correlations in the table are independent of
GDP per capita differences between countries.13 The key correlations include:
11
It should be noted that simultaneous correlation analysis does not provide an estimate of causality. Policy
and environmental variables may drive efficiency, but the reverse may also be true, and unobserved
variables may drive policy and environmental variables as well as efficiency.
12
Afonso and St. Aubyn (2007), using bootstrap procedures to assess the impact of exogenous factors on
the variation of health efficiency across countries, also find that higher GDP levels are associated with
higher system efficiency. Additionally, they find that a high level of education attainment in a country
improves health system efficiency, while the prevalence of obesity and tobacco consumption lower health
system efficiency.
13
Several of the factors that are correlated with relative efficiency are also significantly correlated with
GDP. For instance, countries with higher income levels spend more on pharmaceuticals, have higher out-of-
pocket expenditures, and have better access to medical technology such as MRI equipment. Simultaneous
correlations between these factors and relative efficiency levels may thus simply reflect the strong
association between GDP and the efficiency level. Thus, in order to isolate the effects of the associated
factor on efficiency from its relationship with GDP, in cases where the associated factor is significantly
correlated with GDP, we ran simple regressions of relative efficiency on the associated factor and GDP per
15
These results suggest that changing the mix of real resources is key for improving the
system efficiency of health spending in the Slovak Republic. System efficiency may be
raised from current low levels by containing pharmaceutical costs, doctors’ consultations,
bed utilization, and outpatient contacts, as well as the number of hospital beds.
capita. In those cases, the reported correlations are the regression coefficient of the associated factor, and
are only reported when the coefficient is statistically significant.
14
A World Bank and USAID (2000) study and a report by International Business Strategies (2006) show
that the Slovak health system suffers from corruption and that individuals may be willing to pay for better
health services. However, this is unlikely without an improvement in the quality of health services.
16
Sources: WHO Europe, World Bank World Development Indicators, and the OECD.
1/ Correlations were run on bias-corrected output-oriented efficiency scores. This table summarizes the results of the
correlations of associated factors with the level of efficiency. ++ (+) indicates that the associated factor is positively correlated
with level of efficiency (negatively correlated with output-oriented efficiency scores) at the 5 (10) percent significance level. – –
(–) indicates that the associated factor is negatively correlated with level of efficiency (positively correlated with output-oriented
efficiency scores) at the 5 (10) percent significance level. Several of the associated factors in the table are highly correlated with
GDP. When a factor is correlated with GDP, only correlations that are significant after conditioning on GDP are considered. (See
Appendix I).
2/ Only real health resources/services not included in the DEA (hospital beds, number of physicians, health workers,
pharmacists, and measles immunization rate) are considered.
3/ Excludes non-OECD countries due to missing data.
4/ Excludes the non-European countries Australia, Canada, Japan, Korea, New Zealand, and the U.S. due to missing data.
17
The immediate challenge for the health care system in the Slovak Republic is to improve
health sector outcomes while containing public health spending. Medium-term fiscal
consolidation objectives imply limited room for increasing health spending. At the same
time, health spending may come under pressure from demands for wage increases and
rising prices for pharmaceuticals. Therefore, in order to prevent a deterioration in the
financial condition of health care institutions and achieve further gains in health
outcomes, the efficiency of spending will need to be increased.
In order to meet this challenge and raise the efficiency of health spending, the mix and
quality of real resources needs to be improved. Like other NMS-10 countries, the Slovak
Republic has relatively high cost effectiveness, but low system efficiency. System
efficiency may be enhanced by containing the cost of pharmaceuticals and reducing the
reliance on hospital care. In addition, spending efficiency can be raised through higher
out-of-pocket expenditure and more cost-effective administrative arrangements.
Introduction of the right incentives will be critical for improving health care spending
efficiency. The Slovak health care system is decentralized, and the central government
has limited control over decisions by insurance companies and health care institutions.15
Therefore, a successful framework for health reform needs to include incentives for
implementation, together with enhanced transparency and improved accountability.
The following measures could contribute to raising efficiency and containing health costs:
• Reduce the reliance on hospitals and contain the cost of hospital care. This could
involve various actions:
15
Large hospitals connected to universities are still under central government control and are the main
exception.
18
• Reintroduce co-payments for doctors’ visits and hospital care. Containing the
number of doctors’ visits and prescriptions would help contain the consumption of
pharmaceuticals. The co-payments for hospital stays may help to optimize the
utilization of hospital beds, which are an abundant resource. They may also
reduce the in-care admissions rate, which may also help increase health system
efficiency.
• Increase the power of the antitrust authority and enhance the autonomy and
independence of the Health Care Supervisory Board. Tight supply and
information asymmetries often hinder effective competition in health
service provision (OECD, 2003). Partly because both insurance companies
and hospitals are still largely government-owned, and partly because of the
special relationship between patient and doctor (based on information
asymmetries and trust), competition in the health sector is structurally
lower than in other sectors of the economy. The authorities should closely
monitor the sector (including the Pharmaceutical Reimbursement
Commission), sanction anticompetitive and unethical behavior (such as
collusion among public sector agencies or in the relationship between
pharmaceutical companies and health care providers), and enhance
transparency. The Health Care Supervisory Board should become
independent from the ministry of health (which is directly or indirectly a
supervised institution), and appointment to the Board should be shifted
from government to parliament or to the president.
This annex discusses the cross-country empirical methodology used to investigate the
relationship between spending and outcomes in the health sector. A second-stage
statistical analysis is used to assess whether differences between countries can be
accounted for by factors out of the control of policy makers, or whether changes in
expenditure and sectoral policies may impact the link between health inputs and
outcomes.
The framework of production efficiency can be used to assess the relative efficiency with
which production units convert input items into production items (i.e., technical
efficiency). In Figure 4, as production unit A achieves the same or more product items as
production unit E with fewer input items, unit A is more efficient16 than unit E. Similarly,
unit E is less efficient than units B, C, and D. The difference between the input items used
by units A and E can be used to measure the inefficiency of unit E relative to unit A.
(Alternatively, this could be measured by the difference in production items.)
The most efficient units in a sample provide the parameters for an initial estimate of the
best-practice frontier. One of the most common ways for determining the best-practice
(or production possibility) frontier is DEA (a more detailed discussion of DEA can be
found in Zhu, 2003). The best-practice frontier is illustrated in Figure 4 by the solid line
that connects the best-practice units A, B, C, and F. Because these are the most efficient
units in the sample, they are assigned an efficiency score of 1. The efficiency scores of
the less efficient units (D and E) depend on their distance to the best-practice frontier.
Several measures of the distance to the frontier can be used. Here we adopt the Farrell
input efficiency score.
16
We use the terms (in)efficient and (in)efficiency to describe the link between input items and production
(outcome) items, as the link not only captures strengths or weaknesses in the production or system process,
but is also influenced by uncontrollable/non-discretionary associated factors.
21
B
2
Product item
1.5
D
A
1
E
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Input item
In this paper, we only assess cases with one input and one production item and focus on
output-oriented efficiency. In this simple case, it is straightforward to calculate the
efficiency score for unit E as the ratio of the number of inputs needed at a minimum
(i.e., at the frontier) to achieve its level of production and the number of inputs actually
used by unit E. The score, called the input-oriented efficiency score, can be interpreted as
an indicator of the cost savings that could be achieved from efficiency enhancement.
Alternatively, the output-oriented efficiency score for unit E can be calculated as the ratio
of the number of outputs achieved at a maximum (i.e., at the frontier) to the number of
input items actually used by unit E. The output-oriented efficiency score reflects the
improvement in outputs that could be achieved from efficiency enhancement. As the
Slovak Republic in the immediate term will need to improve health outcomes without
increasing expenditures, we focus only on output-oriented efficiency scores.
Simple DEA estimation produces biased estimates of the efficiency scores which need to
be corrected. Estimating the best-practice frontier for the sample of countries from the
observations of health spending and outcomes is subject to bias, for which a correction
needs to be made. This bias stems from the fact that since we only observe a subsample of
the possible outcomes representing all feasible combinations of spending and outcomes,
we do not know the exact position of the best-practice frontier. Suppose, for example, that
the inputs in Figure 4 represent health spending and the production years of health-
adjusted life expectancy. Also, suppose that health spending or years of health-adjusted
life expectancy were initially not observed for country B. Then the best-practice frontier
would be drawn through the countries A, C, and F, following the dashed line between the
observations for countries A and C. However, suppose the observation for country B
becomes available. Then the best-practice frontier would shift outward to the line that
connects countries A, B, C, and F. It is straightforward to see that, as a general principle,
22
as more information becomes available about the feasible production combinations, the
best-practice frontier may shift outward but cannot move inward. This one-sided error
means that estimating the best-practice frontier with a finite sample is subject to bias.
Since output–oriented efficiency scores are measured in relation to the frontier, the
estimated scores are subject to the same finite sample downward bias (i.e., the level of
efficiency is overestimated unless a correction is made for the bias).
Corrections are made for the estimation bias in the best-practice frontier and efficiency
scores through bootstrapping. This paper uses a method proposed by Simar and Wilson
(2000), and is based on the assumption that the frontier that envelops all possible
production combinations of input and product items is smooth. A key issue is how
quickly the estimated efficiency scores converge to their unbiased true values if the
sample of observations is expanded.17 In the case of one input and one production item,
the convergence rate is fast enough to yield acceptable estimates of efficiency scores and
build confidence intervals. The bootstrapping routine was run in FEAR, a software
package developed by Paul W. Wilson (Wilson, 2006). Table 7 in the main text presents
rankings for the NMS-10 countries relative to the OECD and EU sample based on the
point estimates of the bias-corrected efficiency scores.
The analysis of the link between spending and outcomes in health is complicated by the
fact that this relationship is indirect. Spending has no direct impact on outcomes. Rather,
spending translates into real resources (e.g., hospital beds and service delivery contracts
with physicians) which are combined to produce intermediate outputs
(e.g., immunizations rate and patient-doctor consultations). These real resources in turn
are used to promote better outcomes.
The DEA results can be disaggregated to assess at what stage of the spending process
inefficiencies arise. In particular, as shown in Figure 2, the analysis attempts to
disaggregate what happens in the stage from spending to intermediate outputs (cost
efficiency) and from intermediate outputs to outcomes (system efficiency). This is done
by comparing spending efficiency (the overall measure of efficiency from spending to
outcomes as discussed above) and system efficiency (see Table 8). First, an index of
intermediate inputs (by level) is created. The intermediate output indicator is the index of
the countries’ average ranks for number of hospital beds, immunizations, physicians,
health workers, and pharmacists per capita. Second, efficiency scores are calculated,
using the intermediate output index as an input and various associated outcomes (infant,
child, and maternal mortality rates, as well as HALE, standardized death rates and the
17
This convergence speed is n-2/(p+q+1), where p is the number of inputs and q is the number of production
items. In the 1 input / 1 product cases of this paper, the convergence speed is n-2/3. This is faster than the
convergence speed for a standard parametric regression of n-1/2, suggesting that reasonable estimates of
efficiency scores and confidence intervals can be reached with a lower number of observations than would
be needed for standard regression analysis. However, the convergence speed declines exponentially as the
number of inputs and production items is increased, and already at two inputs and production items, the
speed of convergence is markedly slower than for parametric regression. This implies that such an
expansion in numbers of inputs and production items comes at great cost in terms of the ability to draw
conclusions about efficiency from a limited number of observations.
23
incidence of tuberculosis). Third, the resulting system efficiency rankings are averaged,
expressed as a ratio of the average OECD ranking, and compared with similar ratios for
spending efficiency.
Simple pair-wise correlations are used to determine which policy factors and other factors
outside of the immediate control of policymakers influence the link between health
expenditures and outcomes. Due to the small sample size, we refrain from using
regression analysis in the second stage. With larger sample sizes, regression analysis
provides insight into the relative importance of factors that are associated with efficiency.
The standard approach for the second stage is to regress the DEA efficiency scores on a
set of explanatory variables. However, direct estimation yields biased coefficient
estimates due to serial correlation between the observations (the source of this is the same
as the bias in the estimation of efficiency scores discussed above) which can be corrected
by another set of bootstrap procedures proposed and developed by Simar and Wilson
(2007) and applied by Afonso and St. Aubyn (2006).
Given the close relationships of spending and outcomes with income levels, correlations
of efficiency scores and associated factors are conditional on GDP. GDP per capita has a
very strong negative impact on efficiency. Many of the factors that are associated with
efficiency are also closely related to income level. In order to avoid attribution of factors
whose effects on the variation in efficiency cannot be separated from the effect of GDP,
only GDP per capita and factors that are correlated with efficiency independently of GDP
per capita are considered in the second-stage analysis of this paper. The association with
efficiency of factors that are strongly correlated with GDP is assessed by regressing the
efficiency score on both GDP and the associated factor.
24
REFERENCES
Afonso, António, and Miguel St. Aubyn, 2006, “Cross-Country Efficiency of Secondary
Education Provision: A Semi-Parametric Analysis with Non-Discretionary
Inputs,” Economic Modelling, Vol. 23 (May), No. 3, pp. 476–91.
———, 2007, “Assessing Health Efficiency Across Countries with a Two-step and
Bootstrap Analysis,” mimeo.
Health Policy Institute, 2007, “Basic Health Policy Framework for 2007–2010.”
Available via the Internet: www.hpi.sk.
Institute for Health Information and Statistics, 2005, “Health Statistics Yearbook of the
Slovak Republic 2004,” Bratislava. Available via the Internet: www.uzis.sk.
International Business Strategies, 2006, “Health Care Services in Slovakia.” Available via
the Internet: www.internationalbusinessstrategies.com.
Mattina, Todd, and Victoria Gunnarsson, 2007, “Budget Rigidity and Expenditure
Efficiency in Slovenia,” IMF Working Paper 07/131 (Washington: International
Monetary Fund).
Nemec, Juraj, and František Ochrana, 2005, “Slovak Health Care Reform and its Impact
on Public Finance,” In Macro-Fiscal Policies: New Perspectives and Challenges:
Congress of the IIPF, Korea, 2005, S: 94.
OECD, 2003, “Health-Care Systems: Lessons from the Reform Experience,” Economics
Department Working Paper No. 374. Available via the Internet:
http://www.oecd.org/dataoecd/5/53/22364122.pdf
———, 2006, “The Supply of Physician Services in OECD Countries,” Directorate for
Employment, Labour and Social Affairs Group on Health, OECD Health Working
Papers No. 21 (DELSA/HEA/WD/HWP(2006)1. Available via the Internet:
http://www.oecd.org/dataoecd/27/22/35987490.pdf
Simar, Léopold, and Paul W. Wilson, 2000, “A General Methodology for Bootstrapping
in Non-Parametric Frontier Models,” Journal of Applied Statistics, Vol. 27, No. 6,
pp. 779–802.
25
Wilson, Paul, W., 2006, “FEAR: A Software Package for Frontier Efficiency Analysis
with R. Available via the Internet:
www.clemson.edu/economics/faculty/wilson/courses/bcn/papers/fear.pdf.
Zhu, Joe, 2003, Quantitative Models for Performance Evaluation and Benchmarking
(Boston: Kluwer Academic Publishers, Vol. 51).